JHU ASpIRE system: Robust LVCSR with TDNNS, iVector adaptation and RNN-LMS

By Vijayaditya Peddinti, Guoguo Chen, Vimal Manohar, Tom Ko, Daniel Povey, Sanjeev Khudanpur. 2015

115 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Model Architectures, Speech & Audio, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Model Architectures

In Speech & Audio

In Signal Processing

Citations in the corpus

VOXLINGUA107: A Dataset for Spoken Language Recognition (2020, methodology): “Since the audio in LRE07 evaluation data contains a lot of non-speech regions, we removed long regions of non-speech us-ing a Kaldi speech activity detection model trained for the IARPA ASpIRE challenge by the JHU team [23], prior to extracting embeddings.”
ContextNet: Improving Convolutional Neural Networks for Automatic Speech Recognition with Global Context (2020, background): “Previous works on hybrid ASR have successfully introduced the context to acoustic models by either stacking a large number of layers, or having a separately trained global vector to represent the speaker and the environment information [13, 14, 15, 16].”
Wav2Letter: an End-to-End ConvNet-based Speech Recognition System (2016, background): “Recently, [23] proposed GMM-free training, but the approach still requires to generate a force alignment.”
Letter-Based Speech Recognition with Gated ConvNets (2017, background): “We detail each of these components in the following.”
Sentiment Analysis using Imperfect Views from Spoken Language and Acoustic Modalities (2018, methodology): “To obtain the ASR view, we use the public domain Kaldi ASR toolkit (Povey et. al., 2011) along with the ASpIRE Chain acoustic models (Peddinti et al., 2015; Povey, 2017).”
On Multi-Domain Training and Adaptation of End-to-End RNN Acoustic Models for Distant Speech Recognition (2017, background): “Such multi-domain trained DNN acoustic models provide very competetive accuracies, often outperforming most other approaches in benchmarks [6].”
The Automatic Speech recogition In Reverberant Environments (ASpIRE) challenge (2015, methodology): “Team B [23]: System 14, the single microphone system submitted by Team B, was a multi-splice DNN system [24].”
A study on data augmentation of reverberant speech for robust speech recognition (2017, methodology): “reported a 33% relative improvement from data augmentation[4].”
Mitigating Noisy Inputs for Question Answering (2019, methodology): “We then decode these utterances using the Kaldi chain model [7], trained on the Fischer-English dataset [8] for consistency with past results on mitigating ASR errors in MT [6].”
End-to-End Speaker Diarization for an Unknown Number of Speakers with Encoder-Decoder Based Attractors (2020, methodology): “In these cases, TDNNbased speech activity detection [38] and the oracle number of speakers were used for the evaluation.”
Multimodal Abstractive Summarization for How2 Videos (2019, methodology): “We use the state-of-the-art models for distant-microphone conversational speech recognition, AS-pIRE (Peddinti et al., 2015) and EESEN (Miao et al., 2015; Le Franc et al., 2018).”
Acoustic Modelling from the Signal Domain Using CNNs (2016, background): “However, iVectors are not straight-forward to work with, especially in mismatched conditions [17], and requires careful pre-processing such as segmentation and additional architectural tricks [18].”